# How to Choose a Four-Card Keno Layout Without Mistaking It for an Edge

> Choose a four-card keno paytable and total stake, then compare layouts. See why overlap changes results—not RTP—and how to check payouts.

- Source: https://toolsgambling.com/blog/4-card-keno-strategy
- Author: Evgeniy Volkov (ToolsGambling)
- Published: 2026-02-23
- Updated: 2026-10-14
- Cite as: "4 Card Keno Strategy: Paytables, Overlap and Round Cost", ToolsGambling, https://toolsgambling.com/blog/4-card-keno-strategy

The practical approach to four-card keno is to choose the paytable and total round stake first, then choose a layout for the distribution of results you prefer. Overlapping numbers can make payouts arrive together; spreading numbers across cards can change how often at least one card pays. Neither overlapping numbers nor a visual pattern creates an edge. There is no universal best set of numbers or spot count.

The calculations below use a **hypothetical four-spot paytable**, not a current casino or manufacturer table. They show how three layouts can have the same expected return but different round outcomes.

## What four-card keno actually means

Four-card keno uses up to four separately paid cards, commonly labeled A, B, C and D, settled against one shared draw. Four cards are not four spots: a spot is one selected number on a card. A four-card game can therefore have four spots on each card, subject to its rules.

In the classic illustrative model, the game draws 20 different numbers from a pool of 80, without replacement. Each selected number has a 25% chance of appearing. Those 20 numbers are not 20 independent selections with replacement.

Successive draws may be independent under a fair model, but the four cards within one draw are dependent—even when their selected numbers do not overlap. Four separately paid cards are not four independent trials.

A historical description of this format appears in [US6315291B1](https://patents.google.com/patent/US6315291B1/en), published November 13, 2001. It describes A/B/C/D cards with identical, different or overlapping selections, separate stakes and one common 20-number draw. It establishes a historical format, not current operator rules or a modern machine paytable.

![Original English text excerpt from the historical patent description about four keno cards settled by a common 20-number draw.](https://toolsgambling.com/media/article/2026/10/b8c8b9b6d8cb3cab6158f6ac36e3fb5d.webp)

*Historical source: US6315291B1, published November 13, 2001. This original English text excerpt from the historical patent description is not a game-interface illustration or a current game paytable.*

### How to set up a round

Start by confirming the actual four-card mode, its rules and its full cost before making selections. Some games display four card slots while allowing fewer than four to be active.

Use the controls and rules available in your game to:

1. **Find the correct mode.** Confirm that it settles multiple cards against a shared draw.
2. **Check the number pool and draw size.** Do not assume every variant uses 80 numbers and a 20-number draw.
3. **Check active cards and allowed spots.** Confirm which cards will be charged and whether different spot counts are permitted.
4. **Select numbers on each active card.** Repeated numbers across cards are allowed only where the game permits them.
5. **Open the relevant paytable.** Check the table for the selected spot count, denomination and any feature.
6. **Establish the gross return.** Determine what each listed award returns relative to that card's stake.
7. **Set the total cost.** For equal stakes, calculate denomination × credits per card × active cards. If separate stakes are allowed, add the individual card costs instead.
8. **Inspect the final setup.** Check numbers, stakes, active cards and feature settings before playing.

There is no universal interface label for these actions. Use the game's actual controls rather than assuming that another machine's buttons or rules apply.

## Which numbers and spot counts should you choose?

No particular numbers improve a fair card's probabilities when the spot count stays the same. Shapes, repeated selections, lucky numbers, hot or cold numbers, and a random-selection feature all have the same single-card catch probabilities for an equal number of spots.

Spot count does matter because it changes the catch distribution and determines which paytable applies. That does not make four, five or six spots universally best. Compare the actual tables offered by the game, rather than choosing a spot count from a general rule.

Mixed spot counts are possible only if supported. A four-spot card and a six-spot card need separate paytable assessments; the four-spot calculations below cannot price both.

Keeping or changing numbers between fair independent draws also does not improve the next draw's odds. The [gambler's fallacy guide](https://toolsgambling.com/casino/gamblers-fallacy) explains why a number being absent recently does not make it due.

## What does the hypothetical four-spot paytable return?

This hypothetical paytable has an RTP of **88.84%** and a house edge of **11.16%** under the classic 80/20 model. It pays something on approximately **25.89%** of cards, but an individual card makes a profit only when it catches three or four numbers.

A **gross payout multiplier** includes the stake returned. A 1× payout returns that card's stake; it is not a profit equal to the stake.

| Catches on a four-spot card | Probability | Hypothetical gross payout multiplier |
|---|---:|---:|
| 0 | 30.83% | 0× |
| 1 | 43.27% | 0× |
| 2 | 21.26% | 1× |
| 3 | 4.32% | 5× |
| 4 | 0.31% | 150× |

The displayed probabilities are rounded to two decimal places. Calculations use the unrounded probabilities: 30.832142541%, 43.273182514%, 21.263546580%, 4.324789135% and 0.306339230%, respectively.

For four spots, the mean number of catches is one. Catching all four has probability approximately 1 in 326.44. That is an average probability, not a promise of one occurrence every 326 draws.

### How the probabilities are calculated

The correct single-card calculation is the hypergeometric distribution because the game draws different numbers without replacement. For a card with $n$ selected numbers:

$$
P(K=k)=\frac{\binom{n}{k}\binom{80-n}{20-k}}{\binom{80}{20}}
$$

Here, $K$ is the number of catches and $\binom{a}{b}$ means the number of ways to choose $b$ items from $a$ items. The numerator counts draws containing exactly $k$ selected numbers and $20-k$ unselected numbers. The denominator counts all possible 20-number draws from 80.

The [official SciPy hypergeometric documentation](https://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.hypergeom.html) gives the underlying without-replacement distribution.

For this hypothetical paytable:

$$
RTP=\sum_{k=0}^{4}P(K=k)\times\text{gross payout multiplier}(k)
=88.838376813\%\approx88.84\%
$$

The house edge is therefore approximately 11.16%. The probabilities of two, three and four catches sum to 25.89%, the probability of any payout. Only three or four catches produce a profit on that individual card, giving a card-profit probability of 4.63%.

## How does overlap change four-card results?

Overlap changes the distribution of the combined result, not the RTP, when all four cards use the same spot count, paytable and stake. In this example, spreading selections raises the probability of at least one payout, while identical cards concentrate all four payouts on the same catch count.

Each card below has four spots and stakes **\$0.25**, for a total round cost of **\$1**. All use the hypothetical table above.

| Layout | Selected numbers |
|---|---|
| Identical | A = B = C = D = {1, 2, 3, 4} |
| Shared core of two | A = {1, 2, 3, 4}; B = {1, 2, 5, 6}; C = {1, 2, 7, 8}; D = {1, 2, 9, 10} |
| Disjoint selections | A = {1, 2, 3, 4}; B = {5, 6, 7, 8}; C = {9, 10, 11, 12}; D = {13, 14, 15, 16} |

“Any payout” means at least one card returns money. A “profitable round” means the combined gross return exceeds the full \$1 cost.

| Layout | Any payout probability | Profitable round probability | RTP | Standard deviation of one-round net result |
|---|---:|---:|---:|---:|
| Identical | 25.89% | 4.63% | 88.84% | \$8.33 |
| Shared core of two | 53.46% | 13.40% | 88.84% | \$4.57 |
| Disjoint selections | 72.19% | 17.63% | 88.84% | \$4.14 |

These results come from exact weighted enumeration of groups of drawn numbers using a multivariate hypergeometric calculation. They are not simulated play, testing results or observed win frequencies. The joint weights sum to one, and the mean return independently agrees with the stated RTP.

Standard deviation measures the spread of one-round net results, in dollars. It is not the maximum loss, a bankroll requirement or a probability of bankruptcy. The ordering in this table should not be generalized to every paytable, bonus feature or risk measure.

The [Wizard of Odds analysis of multi-card variance](https://wizardofodds.com/games/keno/multi-card-variance/), updated August 3, 2026, explains the role of covariance and variance. The figures here are our separate four-spot model, not that source's six-spot table.

### Why you cannot treat disjoint cards as independent

Disjoint selections still share a finite draw, so their payout events are not independent. Using the single-card payout probability in the shortcut $1-(1-p)^4$ gives the wrong four-card answer.

With $p=25.8947\%$, that shortcut produces **69.84%**, compared with the exact **72.19%** any-payout probability for the disjoint layout above.

Equal expected value can coexist with different payout probabilities, profit probabilities and variance. There is no universal best overlap count.

## When does a payout actually mean a profit?

A payout means a card returned money; a profitable round means all card returns together exceeded the total stake. A catch alone does not necessarily produce either.

For example, one catch pays nothing in the hypothetical table. Two catches return the individual card's stake. If one disjoint card catches two numbers and the other three pay nothing, the combined return is \$0.25 against a \$1 cost: a **\$0.75 net loss**.

If all four identical cards catch two numbers, each returns \$0.25. Their combined \$1 return breaks even for the round.

### A constructed draw showing how payouts stack

A chosen draw can demonstrate settlement, but it cannot establish which layout is more profitable over time. Consider this explicitly constructed 20-number draw—not a played win:

$$
\{1,2,3,4,17,18,\ldots,32\}
$$

It contains the four numbers 1–4 plus the sixteen numbers 17–32.

| Layout | Catches on A / B / C / D | Combined gross return | Net round result |
|---|---|---:|---:|
| Identical | 4 / 4 / 4 / 4 | \$150.00 | +\$149.00 |
| Shared core of two | 4 / 2 / 2 / 2 | \$38.25 | +\$37.25 |
| Disjoint selections | 4 / 0 / 0 / 0 | \$37.50 | +\$36.50 |

To repeat the calculation, multiply each card's \$0.25 stake by its applicable gross multiplier, add the four returns, and subtract the \$1 total cost once.

For identical cards, each returns \$0.25 × 150 = \$37.50, totaling \$150. For the shared-core layout, card A returns \$37.50 and the other three return \$0.25 each. For disjoint selections, only A pays.

Identical cards do best on this deliberately chosen draw. That conditional result is not evidence that identical cards are the best strategy.

## How to check the paytable with the calculator

Use the [keno calculator](https://toolsgambling.com/casino/keno-calculator) to evaluate the single-card paytable before thinking about four-card layouts. It uses a fixed 80-number pool and 20-number draw, supports one through ten spots, and updates results automatically.

To reproduce the hypothetical example:

1. Select **four spots**.
2. Enter the payouts for catches **0, 1, 2, 3 and 4** as **0, 0, 1, 5 and 150**.
3. Read the automatically updated results. There is no “Calculate” button.
4. Check **RTP 88.84%**, **house edge 11.16%**, **chance to win anything 25.89%** and **average catches 1.00**.
5. Change only the four-catch multiplier from **150 to 100**. RTP falls to **73.52%**, while catch probabilities stay the same.
6. Restore the four-catch multiplier to **150**.

This comparison isolates the paytable's effect. Favorite numbers cannot compensate for a smaller award.

![English-language local keno calculator screenshot with the hypothetical four-spot gross multipliers 0, 0, 1, 5 and 150.](https://toolsgambling.com/media/article/2026/10/d15a7162974cab7c1c85bf23577e13e1.webp)

*English-language screenshot captured locally using synthetic inputs, not a production wager. “Best spot” compares stored example rows; it is not a validated recommendation for a real game's unentered paytables.*

For a real table, enter every paying catch count, including zero if it pays. If an award is explicitly quoted as profit only, convert it to gross return by including the returned stake. A fixed-credit award must be normalized by the actual per-card stake. Caps and bonuses require their exact rules.

The calculator does **not** model four-card layouts, sessions, bonus features or progressive awards. Its “Best spot” comparison uses stored examples for other spot counts. Unless every relevant actual spot table has been entered, it cannot establish the best spot count for your game.

## What can the pattern visualizer show?

The pattern visualizer displays predefined selections so you can inspect how cards share or separate numbers. It does not simulate draws, calculate odds or RTP, or price the three four-spot layouts above.

Switch among its three tabs and inspect the selected numbers for each card:

- **“Quadrant Split”** uses five numbers per card in numerical bands: 1–20, 21–40, 41–60 and 61–80. These are not geometric quarters of the screen.
- **“Overlapping Numbers”** uses six numbers per card with varying shared numbers. It is not the shared-core-of-two four-spot model calculated above.
- **“Column Cluster”** uses five numbers per card.

For example, the “Quadrant Split” selections are A: 3, 7, 11, 15, 19; B: 23, 28, 32, 36, 40; C: 42, 47, 51, 56, 60; and D: 63, 68, 72, 76, 80.

Do not apply the four-spot payout table to these five- or six-spot patterns. The statement that number selection does not affect RTP applies when spot count and paytable stay the same. Changing either requires a new calculation.

## How much should you budget for four cards?

Set an affordable total round stake and loss limit before dividing money among cards. Four cards do not inherently cost more than one card; they cost more when you add cards without reducing the stake on each.

With four cards at \$0.25 each, the round costs \$1. Under this hypothetical 88.84% RTP model:

| Planned rounds | Total wagering turnover | Model expected loss |
|---|---:|---:|
| 100 | \$100 | \$11.16 |
| 200 | \$200 | \$22.32 |

Turnover is the sum of stakes placed, not necessarily the amount deposited. Expected loss is a long-run mathematical average, not a session forecast, guaranteed loss limit or required starting bankroll.

One card staking \$1 on the same gross paytable has the same expected return as four cards staking \$0.25 each. Four identical \$0.25 cards also have the same full round-result distribution as one \$1 card if payouts and limits are linear.

By contrast, keeping the usual \$0.25 card stake and increasing from one card to four raises the round cost from \$0.25 to \$1. Under identical linear mechanics, 20 cards at \$0.25 would cost \$5 per draw rather than \$1 for four cards. That is greater exposure, not automatically a worse RTP.

A minimum denomination may prevent splitting your intended budget across four cards. Reduce the active-card count or pass rather than increasing spending to satisfy the layout.

Use fixed affordable loss and time limits, without chasing losses or increasing stakes through a martingale. A simple log can record the paytable, spot counts, layout, card stakes, number of rounds and combined returns. The [bankroll management guide](https://toolsgambling.com/blog/bankroll-management-guide) provides further budgeting context; no bankroll system removes the house edge.

## What changes in bonus variants or demo play?

Bonus variants need their own rules and paytables; the plain 80/20 model cannot price extra balls, multipliers or progressive awards by name alone. The same brand can offer different tables at different denominations.

For games with Cleopatra- or Caveman-style extras, obtain the exact feature description and matching paytable before estimating return. Do not enter only the base awards into the plain single-card tool and treat the output as the full game's RTP.

Demo play can help identify controls, active-card settings and how awards are displayed. It does not prove future wins, establish RTP from a short run or validate a layout's profitability.

## What should you check before playing?

Choose the game only after you understand its paytable and total cost; choose the layout afterward. The final check is whether the rules, stakes and result distribution match what you intend—not whether the numbers look promising.

- **Rules:** Confirm the pool, draw size, active cards, allowed spots and any features.
- **Paytable:** Check gross awards for every paying catch count and the chosen denomination.
- **Cost:** Add all active-card stakes and confirm the displayed round cost.
- **Layout:** Decide whether you want more synchronized results or a different distribution of payouts, without treating either as an edge.
- **Limits:** Set affordable loss and time limits before starting.
- **Expectations:** Separate catches, paid cards, break-even returns and profitable rounds.

The useful four-card strategy is disciplined selection of paytable, cost and layout. Overlap can change how results arrive; it cannot turn an unfavorable paytable into a favorable one.

**Video Keno RTP by Spot Count**

|  | Typical RTP % |
|---|---:|
| 1 spot | 75 |
| 2 spot | 72 |
| 3 spot | 74 |
| 4 spot | 86 |
| 5 spot | 88 |
| 6 spot | 90 |
| 7 spot | 91 |
| 8 spot | 92 |
| 9 spot | 90 |
| 10 spot | 87 |

## FAQ

### Are four cards the same as four spots in keno?

No. Four cards are four selections settled against a shared draw. Four spots means four selected numbers on one card. Each of four cards can have four spots, or other permitted spot counts.

### Should I keep the same numbers or change them between draws?

Under a fair model with independent successive draws, neither choice improves the next draw's probabilities for the same spot count. Lucky numbers, hot or cold numbers, shapes and random selections do not create an edge.

### Can I use different stakes or spot counts on the four cards?

Only if the game allows it. Check each card's actual paytable and stake, then add the card costs to find the total round stake. Overall RTP is stake weighted when card stakes or paytables differ.

### Why can a paid card still produce a losing round?

A card payout must be compared with the cost of all active cards. If four cards cost $1 in total and one returns $0.25 while the others return nothing, the round loses $0.75.

### Does the classic 80-number, 20-number-draw model cover bonus keno?

Not necessarily. Additional balls, multipliers, progressive awards and other features require the exact rules and paytable. The single-card calculator does not model those features or four-card layouts.